thetradingparrot
patreon
QFL Pine script available - backtest and alerts compatible
🕑 Added 2021-06-21 16:31:05 +0000 UTCUpdate 1 - Added bot and QFL settings used in example with MATIC
For a while now I've been trying to find a decent implementation of QFL that we can use for:
- backtesting using the backtest script
- for sending alerts from trading view to DCA bots that currently don't support QFL
- for creating strategies that combine multiple factors, for example: QFL + RSI + Stochastic. Having a pine script that converges all these conditions and sending a BUY signal only when all conditions are met.
Last weekend I looked into a few alternative open-source options available on trading view and decided to give them a try. One was offering multi-time frame but I decided to go based on Quick Fingers Luc's advice on this video:
https://www.youtube.com/watch?v=J3CMHqfE2-c&ab_channel=QuickfingersLuc
and also based on what 3Commas does, which I find pretty good already in terms of results (Bakker and Johano use this):
New QFL script
Luckily I manage to find a decent version of QFL that is single timeframe. I had to migrate it to Pine Script 4 and add support for backtesting and alerts, deal with some repainting issues. The original script is from 2016 and it's got almost 7k likes, but it still needed some clean up.
In the chart below you can see it in action on MATIC/USDT.
The results are very impressive. Even though in 232 days MATIC lowest drawdown from top has been 72% and this DCA setup uses a tiny tiny max deviation of 4.4%, it's managing to close ALL deals in profit and make a very impressive 265%. In May this token went down 72% and there are no red bags.
During those 232 days the longest time it took to close a deal was 5.8 days.
I made the QFL pine script open source and public on Trading View. (if TV moderators decide to take it down I can update this link, sometimes they do that).
Can you add QFL to backtest simulator mode?
Yes, I'm planning now to add MFI, CCI, ULT and QFL ASAP.
For now if you want to backtest QFL you can add it to the chart and use "indicator" deal start condition mode.
Will this version perform exactly the same as the 3Commas built-in QFL one?
The 3Commas version of QFL is closed-source so there's no way I can guarantee that this version will open the same deals or perform similarly.
They might be using some optimisations or internal parameters that might make it slightly different. A paper trading bot could potentially show that it's taking similar deals though.
What are the settings used in the bot mentioned in the example?
DCA settings:
- BO 10
- SO 20
- 2 SO's
- % dev = 2
- vol scale = 2
- step scale = 1.2
- TP 6%
QFL settings:
- TF = 2 hours
- Vol MA = 1000
- percentage = 4.5
- max candle age = 0
Comments
how do i get access to this backtesting script? my tradingview username is "cosmicjoker"
high timeframes are not recommended for backtesting, 1, 5 or 15 mins is fairly accurate still
I mean will you share this script and we will test it with the currencies we have already, or are you planning to create a specific bot with some pairs already?
Step? For what Marcello?
So what is the next step?
Okey. Well if backtest shows these results im more than happy to try it out. You are right about QFL beeing picky. If you want to trade QFL 100% as quick finger luc intended than there is a certain kind of drop followed by a certain kind of bounce that makes a base qualify as a really strong base. Thats Basicly what i meant with the scripts making some sketchy bases. Dosent mean it won't return to base and make a succsesful trade, but it's not as certain as when you see that rapid panic drop followed buy a nice bounce back to base and beyond. No idea why backtest shows so different results on different timeframes tough🤔
I found this one pretty good, and the clean up was mostly to bring the code up to date to pine script 4, repainting, nothing to do with the strat itself. Actually adding the max candle age was an extra new filter I wanted to have, but that's it. Try it out with backtest and see if its good or not. I'm starting to see that QFL is very picky. some times backtest works amazing in 2H and 6H and horrible in 1H and 4H, why??? no idea yet. Some coins work great, some others is a mess. So I'm still in the process of understanding what to take away from this.
Do you mind explaining what kind of clean up you made to the script? I’ve looked trough all the the qfl scripts on TV earlier and scrapped them all because they draw some really sketchy bases.
This is awesome. Will be a part of my army of QFL bots asap.
I'm so excited, I really want to see how it performs now!
thanks!
DCA settings: BO 10 SO 20 2 SO's % dev = 2 vol scale = 2 step scale = 1.2 TP 6% QFL settings: TF = 2 hours Vol MA = 1000 percentage = 4.5 max candle age = 0
You just won the Internet. Well done!
This is huge!
But what are the actual bot settings?
